subscript 3.0.1 → 5.0.0

package/README.md

@@ -1,11 +1,11 @@
 # <!--<img alt="subscript" src="/subscript2.svg" height=42/>--> sub͘<em>script</em> <!--<sub>SUB͘<em>SCRIPT</em></sub>-->
 
-_Subscript_ is micro-language with common syntax subset of C++, JS, Java, Python, Go, Rust.<br/>
+_Subscript_ is micro-language with common syntax subset of C++, JS, Java, Python, Go, Rust, Swift, Objective C, Kotlin etc.<br/>
 
-* It has well-known syntax
+* Well-known syntax
 * Any _subscript_ fragment can be copy-pasted to any target language
 * It's tiny <sub>![npm bundle size](https://img.shields.io/bundlephobia/minzip/subscript/latest?color=brightgreen&label=gzip)</sub>
-* It's very fast ([see performance](#performance))
+* It's :rocket: fast ([see performance](#performance))
 * Configurable & extensible
 * Trivial to use...
 
@@ -26,14 +26,8 @@ _Subscript_ is designed to be useful for:
 * sandboxes, playgrounds, safe eval
 * custom DSL
 
-[_Jsep_](https://github.com/EricSmekens/jsep) is generally fine for the listed tasks, unless you design a tiny module and prefer to keep dependencies as small as possible.
-_Subscript_ has [2.5kb](https://npmfs.com/package/subscript/3.0.0/subscript.min.js) footprint vs [11.4kb](https://npmfs.com/package/jsep/1.2.0/dist/jsep.min.js) _jsep_, with same or better performance. It also has more open API and generates lispy calltree (compatible with [frisk](https://npmjs.com/frisk)), compared to esprima AST: minimal possible overhead, clear precedence, overloading by context, manual evaluation, debugging, conventional form, one-liner docs:
-
-```js
-import {evaluate} from 'subscript.js'
-
-evaluate(['+', ['*', 'min', 60], '"sec"'], { min: 5 }) // min*60 + "sec" == "300sec"
-``` 
+[_Jsep_](https://github.com/EricSmekens/jsep) is generally fine for the listed tasks, unless you need dependencies as small as possible.
+_Subscript_ has [2.5kb](https://npmfs.com/package/subscript/4.0.0/subscript.min.js) footprint vs [11.4kb](https://npmfs.com/package/jsep/1.2.0/dist/jsep.min.js) _jsep_, with better performance.
 
 
 ## Operators
@@ -53,17 +47,53 @@ Default operators include common operators for the listed languages in the follo
 * `&&`
 * `||`
 
-All other operators can be extended via `parse.binary`, `parse.unary` and `evaluate.operator`.
+All other operators can be extended, see [extension](#extension).
+
+## Evaluation
+
+_Subscript_ parser generates lispy calltree (compatible with [frisk](https://npmjs.com/frisk)), which is compared to esprima AST has:
+
++ minimal possible overhead
++ clear precedence
++ overloading by context
++ manual evaluation and debugging
++ conventional form
++ one-liner docs:
 
+```js
+import {evaluate} from 'subscript.js'
+
+evaluate(['+', ['*', 'min', 60], '"sec"'], { min: 5 }) // min*60 + "sec" == "300sec"
+```
+
+## Extension
+
+Tokens are extensible via `parse.token` list, can be added support of _regex_, _array_, _object_, _interpolated string_ and others.
+Default tokens include:
+
+* `"abc"` strings
+* `1.2e+3` floats
+* `()` expression groups or fn calls
+* `.`, `[]` property access
+
+Operators can be extended via `parse.operator` to add support for any unary/binary/postfix operators, calls, props or chains.
+
+Comments can be added via extending `parse.space`.
+
+For now see justin extension how things can be done.
+
+<!--
 ```js
 import { parse, evaluate } from 'subscript.js'
 
 // add precedences
-parse.binary['=>'] = 10
+// TODO
+// parse.operator['=>'] = 10
 
 // define evaluators
-evaluate.operator['=>'] = ( args, body ) => evaluate(body, args)
-evaluate.operator['|'] = ( a, ...b ) => a.pipe(...b)
+// TODO
+// evaluate.operator['=>'] = ( args, body ) => evaluate(body, args)
+// evaluate.operator['|'] = ( a, ...b ) => a.pipe(...b)
 
 let tree = parse(`
   interval(350)
@@ -73,28 +103,13 @@ let tree = parse(`
 `)
 evaluate(tree, { Math, map, take, interval, gaussian })
 ```
-
-## Extending
-
-By default subscript detects the following tokens:
-
-* `"` strings
-* `1.2e+3` floats
-* `true`, `false`, `null` literals
-* `()` expression groups or fn calls
-* `.`, `[]` property access
-
-Literals can be extended via `parse.literal` dict.
-
-Token parsers are extensible via `parse.token` list, can be added support of _regex_, _array_, _object_, _interpolated string_ and others.
-
-Postfix parsers are applied to parsed tokens and can be used to provide _property chains_, _function calls_, _postfix operators_, _token mapping_, _ternary operators_ and so on. They're extensible via `parse.postfix`.
+-->
 
 
 ## Justin
 
 _Justin_ extension (original [thread](https://github.com/endojs/Jessie/issues/66)) is minimal JS subset − JSON with JS expressions.<br/>
-It adds support for:
+It adds support of:
 
 + `**` binary operator
 + `~` unary operator
@@ -104,8 +119,10 @@ It adds support for:
 + `{...}` Object literal
 + `in` binary operator
 + `;` expression separator
-<!-- + `//, /* */` comments -->
-<!-- + `undefined` literal -->
++ unary word operators
++ `//`, `/* */` comments
++ `true`, `false`, `null`, `undefined` literals
+<!-- + `===`, `!==` operators -->
 <!-- + `?` chaining operator -->
 <!-- + `...x` unary operator -->
 <!-- + strings interpolation -->
@@ -262,8 +279,8 @@ Subscript shows relatively good performance within other evaluators:
 // 1 + (a * b / c % d) - 2.0 + -3e-3 * +4.4e4 / f.g[0] - i.j(+k == 1)(0)
 // parse 30k times
 
-subscript: ~280 ms
-jsep: ~282 ms
+subscript: ~200 ms
+jsep: ~280 ms
 expr-eval: ~480 ms
 jexl: ~1200 ms
 new Function: ~1400 ms
@@ -278,6 +295,6 @@ new Function: ~1400 ms
 * [expression-eval](https://github.com/donmccurdy/expression-eval)
 * [jsep](https://github.com/EricSmekens/jsep)
 * [string-math](https://github.com/devrafalko/string-math)
-
+* [nerdamer](https://github.com/jiggzson/nerdamer)
 
 <p align=center>🕉</p>

package/{src/evaluate.js → evaluate.js}

@@ -12,12 +12,14 @@ evaluate = (s, ctx={}, c, op) => {
   }
   if (s && typeof s === 'string')
     return s[0] === '"' ? s.slice(1,-1)
-          : s[0]==='@' ? s.slice(1)
-          : s in ctx ? ctx[s] : s
+      : s[0]==='@' ? s.slice(1)
+      : s in ctx ? ctx[s] : s
 
   return s
 },
 
+reducer=fn=>(...a)=>a.reduce(fn),
+
 // op evaluators
 // multiple args allows shortcuts, lisp compatible, easy manual eval, functions anyways take multiple arguments
 operator = evaluate.operator = {
@@ -25,13 +27,13 @@ operator = evaluate.operator = {
   '++':a=>++a,
   '--':a=>--a,
 
-  '.':(...a)=>a.reduce((a,b)=>a?a[b]:a),
+  '.':reducer((a,b)=>a?a[b]:a),
 
-  '%':(...a)=>a.reduce((a,b)=>a%b),
-  '/':(...a)=>a.reduce((a,b)=>a/b),
-  '*':(...a)=>a.reduce((a,b)=>a*b),
+  '%':reducer((a,b)=>a%b),
+  '/':reducer((a,b)=>a/b),
+  '*':reducer((a,b)=>a*b),
 
-  '+':(...a)=>a.reduce((a,b)=>a+b),
+  '+':reducer((a,b)=>a+b),
   '-':(...a)=>a.length < 2 ? -a : a.reduce((a,b)=>a-b),
 
   '>>>':(a,b)=>a>>>b,
@@ -51,7 +53,7 @@ operator = evaluate.operator = {
   '|':(a,b)=>a|b,
   '&&':(...a)=>a.every(Boolean),
   '||':(...a)=>a.some(Boolean),
-  ',':(...a)=>a.reduce((a,b)=>(a,b))
+  ',':reducer((a,b)=>(a,b))
 }
 
 export default evaluate

package/justin.js

@@ -1,73 +1,101 @@
 // justin lang https://github.com/endojs/Jessie/issues/66
-import {evaluate, operator} from './src/evaluate.js'
-import {parse, binary, unary, postfix, token, literal,
-        code, char, skip, space, expr} from './src/parse.js'
+import {evaluate} from './evaluate.js'
+import {parse, code, char, skip, expr, operator, err} from './parse.js'
+
+// ;
+operator(';', 1)
+
+// ===, !==
+operator('===', 9)
+operator('!==', 9)
 
 // undefined
-literal['undefined'] = undefined
+const v = v => ({valueOf:()=>v})
+parse.token.splice(2,0, c =>
+  c === 116 && char(4) === 'true' && skip(4) ? v(true) :
+  c === 102 && char(5) === 'false' && skip(5) ? v(false) :
+  c === 110 && char(4) === 'null' && skip(4) ? v(null) :
+  c === 117 && char(9) === 'undefined' && skip(9) ? v(undefined) :
+  null
+)
 
-// '
-token.push((q, qc) => q === 39 ? (qc = char(), index++, qc) + skip(c => c !== q) + (index++, qc) : null)
+// "' with /
+parse.token[1] = (q, qc, c, str) => {
+  if (q !== 34 && q !== 39) return
+  qc = char(), skip(), str = ''
+  while (c=code(), c-q) {
+    if (c === 92) skip(), str += escape[char()] || char(); else str+=char()
+    skip()
+  }
+  return skip(), qc + str + qc
+}
+const escape = {n:'\n', r:'\r', t:'\t', b:'\b', f:'\f', v:'\v'}
 
 // **
-binary['**'] = 16
-operator['**'] = (...args)=>args.reduceRight((a,b)=>Math.pow(b,a))
+evaluate.operator['**'] = (...args)=>args.reduceRight((a,b)=>Math.pow(b,a))
+operator('**', 14)
 
 // ~
-unary['~'] = 17
-operator['~'] = a=>~a
+operator('~', 13, -1)
+evaluate.operator['~'] = a=>~a
 
-// ...
-// unary[1]['...']=true
-
-// ;
-binary[';'] = 1
+// TODO ...
 
 // ?:
-operator['?:']=(a,b,c)=>a?b:c
-postfix.push(node => {
+evaluate.operator['?:']=(a,b,c)=>a?b:c
+operator('?', 3, (node) => {
+  if (!node) err('Expected expression')
   let a, b
-  if (code() !== 63) return node
-  skip(), space(), a = expr(58)
-  if (code() !== 58) return node
-  skip(), space(), b = expr()
-  return ['?:',node, a, b]
+  skip(), parse.space(), a = expr()
+  if (code() !== 58) err('Expected :')
+  skip(), parse.space(), b = expr()
+  return ['?:', node, a, b]
 })
+// operator(':')
 
 // /**/, //
-// comments['/*']='*/'
-// comments['//']='\n'
+parse.space = cc => {
+  while (cc = code(), cc <= 32) {
+    skip()
+    if (code() === 47)
+      // /**/
+      if (code(1) === 42) skip(2), skip(c => c !== 42 && code(1) !== 47), skip(2)
+      // //
+      else if (code(1) === 47) skip(2), skip(c => c >= 32)
+  }
+  return cc
+}
 
 // in
 evaluate.operator['in'] = (a,b)=>a in b
-parse.postfix.unshift(node => (char(2) === 'in' ? (skip(2), ['in', '"' + node + '"', expr()]) : node))
+operator('in', 10, (node) => code(2) <= 32 && [skip(2), '"'+node+'"', expr(10)])
 
 // []
-operator['['] = (...args) => Array(...args)
-token.push((node, arg) =>
-  code() === 91 ?
-  (
-    skip(), arg=expr(93),
-    node = arg==null ? ['['] : arg[0] === ',' ? (arg[0]='[',arg) : ['[',arg],
-    skip(), node
-  ) : null
-)
+evaluate.operator['['] = (...args) => Array(...args)
+// as operator it's faster to lookup (no need to call extra rule check), smaller and no conflict with word names
+operator('[', 20, (node,arg) => !node && (
+  skip(), arg=expr(0,93), skip(),
+  !arg ? ['['] : arg[0] === ',' ? (arg[0]='[',arg) : ['[',arg]
+))
 
 // {}
-binary[':'] = 2
-token.unshift((node) => code() === 123 ? (skip(), node = map(['{',expr(125)]), skip(), node) : null)
-operator['{'] = (...args)=>Object.fromEntries(args)
-operator[':'] = (a,b)=>[a,b]
+parse.token.unshift((cc, node) => (
+  cc === 123 && (skip(), node = map(['{', expr(0,125)]), skip(), node)
+))
+
+operator('}')
+operator(':', 0)
+evaluate.operator['{'] = (...args)=>Object.fromEntries(args)
+evaluate.operator[':'] = (a,b)=>[a,b]
 
 const map = (n, args) => {
-  if (n[1]==null) args = []
+  if (!n[1]) args = []
   else if (n[1][0]==':') args = [n[1]]
   else if (n[1][0]==',') args = n[1].slice(1)
   return ['{', ...args]
 }
 
-
 // TODO: strings interpolation
 
-export { default } from './subscript.js';
+export default parse
 export { parse, evaluate }

package/justin.min.js

@@ -1 +1 @@
-var e,r,n,t=e=>Array.isArray(e)&&("string"==typeof e[0]||t(e[0])),l=(e,r={},n,i)=>t(e)?("string"==typeof(n=e[0])&&(i=u[n]),"function"!=typeof(n=i||l(n,r))?n:n.call(...e.map((e=>l(e,r))))):e&&"string"==typeof e?'"'===e[0]?e.slice(1,-1):"@"===e[0]?e.slice(1):e in r?r[e]:e:e,u=l.operator={"!":e=>!e,"++":e=>++e,"--":e=>--e,".":(...e)=>e.reduce(((e,r)=>e&&e[r])),"%":(...e)=>e.reduce(((e,r)=>e%r)),"/":(...e)=>e.reduce(((e,r)=>e/r)),"*":(...e)=>e.reduce(((e,r)=>e*r)),"+":(...e)=>e.reduce(((e,r)=>e+r)),"-":(...e)=>e.length<2?-e:e.reduce(((e,r)=>e-r)),">>>":(e,r)=>e>>>r,">>":(e,r)=>e>>r,"<<":(e,r)=>e<<r,">=":(e,r)=>e>=r,">":(e,r)=>e>r,"<=":(e,r)=>e<=r,"<":(e,r)=>e<r,"!=":(e,r)=>e!=r,"==":(e,r)=>e==r,"&":(e,r)=>e&r,"^":(e,r)=>e^r,"|":(e,r)=>e|r,"&&":(...e)=>e.every(Boolean),"||":(...e)=>e.some(Boolean),",":(...e)=>e.reduce(((e,r)=>r))},i=l,o=t=>(r=t,e=n=0,f()),a=()=>{for(;d()<=32;)e++},s=(r,t,l,u=3)=>{if(e&&n&&n[3]===e)return n;for(;u;)if(null!=(l=r[t=c(u--)]))return n=[t,l,t.length,e]},f=(r,n=-1)=>{a();let t,l,u,i=d(),o=0,p=e;if(i!==r){for(;p===e&&o<g.length;)l=g[o++](i);if(p===e)(t=s(v))&&(e+=t[2],l=[t[0],f(r,t[1])]);else{if(a(),i=d(),i===r)return l;for(o=0;o<x.length;)(u=x[o](l,i))!==l?(l=u,o=0,a(),i=d()):o++}for(a();(i=d())&&i!==r&&(t=s(b))&&t[1]>n;){l=[t[0],l];do{e+=t[2],l.push(f(r,t[1]))}while(c(t[2])===t[0]);a()}return l}},p=e=>(e=y((e=>e>=48&&e<=57||e>=65&&e<=90||e>=97&&e<=122||36==e||95==e||e>=192)))&&A.hasOwnProperty(e)?A[e]:e,d=()=>r.charCodeAt(e),c=(n=1)=>r.substr(e,n),h=r=>{throw Error(r+" at "+e)},y=(n=1,t=e)=>{if("number"==typeof n)e+=n;else for(;n(d());)++e>r.length&&h("Unexpected end "+n);return e>t?r.slice(t,e):null},g=o.token=[(r,n)=>40===r?(e++,n=f(41),e++,n):null,(r,n,t)=>{if(r=y(t=e=>e>=48&&e<=57)||"",46===d()&&(e++,r+="."+y(t)),r)return(69===(n=d())||101===n)&&(e++,r+="e",(43===(n=d())||45===n)&&(r+=c(),e++),r+=y(t)),parseFloat(r)},(r,n)=>34===r?(n=c(),e++,n+y((e=>e!==r))+(e++,n)):null,p],A=o.literal={true:!0,false:!1,null:null},x=o.postfix=[(r,n,t)=>(46===n?(e++,a(),r=[".",r,'"'+p()+'"']):91===n?(e++,r=[".",r,f(93)],e++):40===n&&(e++,t=f(41),r=Array.isArray(t)&&","===t[0]?(t[0]=r,t):null==t?[r]:[r,t],e++),r),(n,t)=>43!==t&&45!==t||r.charCodeAt(e+1)!==t?n:[y(2),n]],v=o.unary={"-":17,"!":17,"+":17,"++":17,"--":17},b=o.binary={",":1,"||":6,"&&":7,"|":8,"^":9,"&":10,"==":11,"!=":11,"<":12,">":12,"<=":12,">=":12,"<<":13,">>":13,">>>":13,"+":14,"-":14,"*":15,"/":15,"%":15},m=o,w=e=>(e="string"==typeof e?m(e):e,r=>i(e,r));A.undefined=void 0,g.push(((e,r)=>39===e?(r=c(),index++,r+y((r=>r!==e))+(index++,r)):null)),b["**"]=16,u["**"]=(...e)=>e.reduceRight(((e,r)=>Math.pow(r,e))),v["~"]=17,u["~"]=e=>~e,b[";"]=1,u["?:"]=(e,r,n)=>e?r:n,x.push((e=>{let r,n;return 63!==d()||(y(),a(),r=f(58),58!==d())?e:(y(),a(),n=f(),["?:",e,r,n])})),l.operator.in=(e,r)=>e in r,o.postfix.unshift((e=>"in"===c(2)?(y(2),["in",'"'+e+'"',f()]):e)),u["["]=(...e)=>Array(...e),g.push(((e,r)=>91===d()?(y(),e=null==(r=f(93))?["["]:","===r[0]?(r[0]="[",r):["[",r],y(),e):null)),b[":"]=2,g.unshift((e=>123===d()?(y(),e=B(["{",f(125)]),y(),e):null)),u["{"]=(...e)=>Object.fromEntries(e),u[":"]=(e,r)=>[e,r];var B=(e,r)=>(null==e[1]?r=[]:":"==e[1][0]?r=[e[1]]:","==e[1][0]&&(r=e[1].slice(1)),["{",...r]);export{w as default,l as evaluate,o as parse};
+var e,r,t=e=>Array.isArray(e)&&("string"==typeof e[0]||t(e[0])),a=(e,r={},o,l)=>t(e)?("string"==typeof(o=e[0])&&(l=n[o]),"function"!=typeof(o=l||a(o,r))?o:o.call(...e.map((e=>a(e,r))))):e&&"string"==typeof e?'"'===e[0]?e.slice(1,-1):"@"===e[0]?e.slice(1):e in r?r[e]:e:e,o=e=>(...r)=>r.reduce(e),n=a.operator={"!":e=>!e,"++":e=>++e,"--":e=>--e,".":o(((e,r)=>e&&e[r])),"%":o(((e,r)=>e%r)),"/":o(((e,r)=>e/r)),"*":o(((e,r)=>e*r)),"+":o(((e,r)=>e+r)),"-":(...e)=>e.length<2?-e:e.reduce(((e,r)=>e-r)),">>>":(e,r)=>e>>>r,">>":(e,r)=>e>>r,"<<":(e,r)=>e<<r,">=":(e,r)=>e>=r,">":(e,r)=>e>r,"<=":(e,r)=>e<=r,"<":(e,r)=>e<r,"!=":(e,r)=>e!=r,"==":(e,r)=>e==r,"&":(e,r)=>e&r,"^":(e,r)=>e^r,"|":(e,r)=>e|r,"&&":(...e)=>e.every(Boolean),"||":(...e)=>e.some(Boolean),",":o(((e,r)=>r))},l=15,s=(t,a)=>(r=t,e=0,a=c(),e<r.length?p():a.valueOf()),p=(t="Bad syntax")=>{throw Error(t+" `"+r[e]+"` at "+e)},u=(t=1,a=e)=>{if("number"==typeof t)e+=t;else for(;t(f());)e++;return r.slice(a,e)},f=(t=0)=>r.charCodeAt(e+t),i=(t=1)=>r.substr(e,t),c=(e=0,r,t,a,o=0,n,l)=>{for(;(t=s.space())&&(l=h[t]?.(a,e)||!a&&d(t));)a=l;return r&&t!==r&&p("Unclosed paren"),a},v=(s.space=r=>{for(;(r=f())<=32;)e++;return r},s.token=[e=>(e=u((e=>e>47&&e<58||46==e)))&&((69==f()||101==f())&&(e+=u(2)+u((e=>e>=48&&e<=57))),isNaN(e=new Number(e))?p("Bad number"):e),(e,r)=>34==e&&u()+u((r=>r-e))+u(),e=>u((e=>e>=48&&e<=57||e>=65&&e<=90||e>=97&&e<=122||36==e||95==e||e>=192&&215!=e&&247!=e))]),d=(e,r=0,t)=>{for(;r<v.length;)if(t=v[r++](e))return t},h=[],y=(r,t=0,a=0,o,n=r.charCodeAt(0),l=r.length,v=h[n],d=r.toUpperCase()!==r,y)=>(y=l<2?d?e=>f(1)<=32:e=>1:d?e=>i(l)==r&&f(l)<=32:e=>i(l)==r,o=a?a>0?e=>e&&[u(l),e]:a<0?e=>!e&&[u(l),(c(t-1)||p()).valueOf()]:a:a=>{a=[r,a||p()];do{e+=l,a.push((c(t)||p()).valueOf())}while(s.space()==n&&y());return a},h[n]=(e,r)=>r<t&&y()&&o(e)||v&&v(e,r));for(let r=0,t=[",",1,,"|",6,,"||",4,,"&",8,,"&&",5,,"^",7,,"==",9,,"!=",9,,">",10,,">=",10,,">>",11,,">>>",11,,"<",10,,"<=",10,,"<<",11,,"+",12,,"+",l,-1,"++",l,-1,"++",l,1,"-",12,,"-",l,-1,"--",l,-1,"--",l,1,"!",l,-1,"*",13,,"/",13,,"%",13,,".",18,(e,r)=>e&&[u(),e,"string"==typeof(r=c(18))?'"'+r+'"':r.valueOf()],"[",18,r=>(e++,r=[".",r,c(0,93).valueOf()],e++,r),"]",,,"(",18,(r,t)=>(e++,t=c(0,41),e++,Array.isArray(t)&&","===t[0]?(t[0]=r,t):t?[r,t.valueOf()]:[r]),"(",19,(r,t)=>!r&&(++e,t=c(0,41)||p(),++e,t),")",,,];r<t.length;)y(t[r++],t[r++],t[r++]);y(";",1),y("===",9),y("!==",9);var g=e=>({valueOf:()=>e});s.token.splice(2,0,(e=>116===e&&"true"===i(4)&&u(4)?g(!0):102===e&&"false"===i(5)&&u(5)?g(!1):110===e&&"null"===i(4)&&u(4)?g(null):117===e&&"undefined"===i(9)&&u(9)?g(void 0):null)),s.token[1]=(e,r,t,a)=>{if(34===e||39===e){for(r=i(),u(),a="";(t=f())-e;)92===t?(u(),a+=O[i()]||i()):a+=i(),u();return u(),r+a+r}};var O={n:"\n",r:"\r",t:"\t",b:"\b",f:"\f",v:"\v"};a.operator["**"]=(...e)=>e.reduceRight(((e,r)=>Math.pow(r,e))),y("**",14),y("~",13,-1),a.operator["~"]=e=>~e,a.operator["?:"]=(e,r,t)=>e?r:t,y("?",3,(e=>{let r,t;return e||p("Expected expression"),u(),s.space(),r=c(),58!==f()&&p("Expected :"),u(),s.space(),t=c(),["?:",e,r,t]})),s.space=e=>{for(;(e=f())<=32;)u(),47===f()&&(42===f(1)?(u(2),u((e=>42!==e&&47!==f(1))),u(2)):47===f(1)&&(u(2),u((e=>e>=32))));return e},a.operator.in=(e,r)=>e in r,y("in",10,(e=>f(2)<=32&&[u(2),'"'+e+'"',c(10)])),a.operator["["]=(...e)=>Array(...e),y("[",20,((e,r)=>!e&&(u(),r=c(0,93),u(),r?","===r[0]?(r[0]="[",r):["[",r]:["["]))),s.token.unshift(((e,r)=>123===e&&(u(),r=b(["{",c(0,125)]),u(),r))),y("}"),y(":",0),a.operator["{"]=(...e)=>Object.fromEntries(e),a.operator[":"]=(e,r)=>[e,r];var b=(e,r)=>(e[1]?":"==e[1][0]?r=[e[1]]:","==e[1][0]&&(r=e[1].slice(1)):r=[],["{",...r]),A=s;export{A as default,a as evaluate,s as parse};

package/package.json

@@ -1,11 +1,12 @@
 {
   "name": "subscript",
-  "version": "3.0.1",
+  "version": "5.0.0",
   "description": "Microlanguage with common syntax for JS/C++/Python/Rust",
   "main": "subscript.js",
   "type": "module",
   "files": [
-    "src/*",
+    "parse.js",
+    "evaluate.js",
     "subscript.js",
     "subscript.min.js",
     "justin.js",

package/parse.js

@@ -0,0 +1,148 @@
+const PERIOD=46, OPAREN=40, CPAREN=41, CBRACK=93, SPACE=32,
+
+  PREC_SEQ=1, PREC_SOME=4, PREC_EVERY=5, PREC_OR=6, PREC_XOR=7, PREC_AND=8,
+  PREC_EQ=9, PREC_COMP=10, PREC_SHIFT=11, PREC_SUM=12, PREC_MULT=13, PREC_UNARY=15, PREC_POSTFIX=16, PREC_CALL=18, PREC_GROUP=19
+
+
+// current string & index
+let idx, cur
+
+export const parse = (str, tree) => (cur=str, idx=0, tree=expr(), idx<cur.length ? err() : tree.valueOf()),
+
+err = (msg='Bad syntax') => { throw Error(msg + ' `' + cur[idx] + '` at ' + idx) },
+
+skip = (is=1, from=idx) => {
+  if (typeof is === 'number') idx += is
+  else while (is(code())) idx++;
+  return cur.slice(from, idx)
+},
+
+code = (i=0) => cur.charCodeAt(idx+i),
+
+char = (n=1) => cur.substr(idx, n),
+
+// a + b - c
+expr = (prec=0, end, cc, node, i=0, map, newNode) => {
+  // chunk/token parser
+  while (
+    (cc=parse.space()) && (newNode = lookup[cc]?.(node, prec) || (!node && token(cc)) )
+  ) node = newNode;
+
+  if (end && cc !== end) err('Unclosed paren')
+
+  return node
+},
+
+// can be extended with comments, so we export
+space = parse.space = cc => { while (cc = code(), cc <= SPACE) idx++; return cc },
+
+// tokens
+tokens = parse.token = [
+  // 1.2e+3, .5 - fast & small version, but consumes corrupted nums as well
+  (number) => (
+    (number = skip(c => (c > 47 && c < 58) || c == PERIOD)) && (
+      (code() == 69 || code() == 101) && (number += skip(2) + skip(c => c >= 48 && c <= 57)),
+      isNaN(number = new Number(number)) ? err('Bad number') : number
+    )
+  ),
+  // "a"
+  (q, qc) => q == 34 && (skip() + skip(c => c-q) + skip()),
+  // id
+  c => skip(c =>
+    (c >= 48 && c <= 57) || // 0..9
+    (c >= 65 && c <= 90) || // A...Z
+    (c >= 97 && c <= 122) || // a...z
+    c == 36 || c == 95 || // $, _,
+    (c >= 192 && c != 215 && c != 247) // any non-ASCII
+  )
+],
+
+token = (c,i=0,node) => { while(i<tokens.length) if (node = tokens[i++](c)) return node },
+
+// operator lookup table
+lookup = [],
+
+// create operator checker/mapper (see examples)
+// @param op is operator string
+// @param prec is operator precedenc to check
+// @param map is either number +1 - postfix unary, -1 prefix unary, 0 binary, else - custom mapper function
+operator = (op, prec=0, type=0, map, c=op.charCodeAt(0), l=op.length, prev=lookup[c], word=op.toUpperCase()!==op, isop) => (
+  isop = l<2 ? // word operator must have space after
+    !word ? c=>1 : c=>code(1)<=SPACE :
+    !word ? c=>char(l)==op : c=>char(l)==op&&code(l)<=SPACE,
+
+  map = !type ? node => { // binary, consume same-op group
+      node = [op, node || err()]
+      // in order to support literal tokens, we call valueOf any time we create or modify calltree node
+      do { idx+=l, node.push((expr(prec) || err()).valueOf()) } while (parse.space()==c && isop())
+      return node
+    } :
+    type > 0 ? node => node && [skip(l), node] : // postfix unary
+    type < 0 ? node => !node && [skip(l), (expr(prec-1) || err()).valueOf()] : // prefix unary
+    type,
+
+  lookup[c] = (node, curPrec) => curPrec < prec && isop() && map(node) || (prev && prev(node, curPrec))
+)
+
+// ,
+for (let i = 0, ops = [
+  // TODO: add ,, as node here
+  ',', PREC_SEQ,,
+
+  '|', PREC_OR,,
+  '||', PREC_SOME,,
+
+  '&', PREC_AND,,
+  '&&', PREC_EVERY,,
+
+  '^', PREC_XOR,,
+
+  // ==, !=
+  '==', PREC_EQ,,
+  '!=', PREC_EQ,,
+
+  // > >= >> >>>, < <= <<
+  '>', PREC_COMP,,
+  '>=', PREC_COMP,,
+  '>>', PREC_SHIFT,,
+  '>>>', PREC_SHIFT,,
+  '<', PREC_COMP,,
+  '<=', PREC_COMP,,
+  '<<', PREC_SHIFT,,
+
+  // + ++ - --
+  '+', PREC_SUM,,
+  '+', PREC_UNARY, -1,
+  '++', PREC_UNARY, -1,
+  '++', PREC_UNARY, +1,
+  '-', PREC_SUM,,
+  '-', PREC_UNARY, -1,
+  '--', PREC_UNARY, -1,
+  '--', PREC_UNARY, +1,
+
+  // ! ~
+  '!', PREC_UNARY, -1,
+
+  // * / %
+  '*', PREC_MULT,,
+  '/', PREC_MULT,,
+  '%', PREC_MULT,,
+
+  // a.b
+  '.', PREC_CALL, (node,b) => node && [skip(),node, typeof (b = expr(PREC_CALL)) === 'string' ? '"' + b + '"' : b.valueOf()],
+
+  // a[b]
+  '[', PREC_CALL, (node) => (idx++, node = ['.', node, expr(0,CBRACK).valueOf()], idx++, node),
+  ']',,,
+
+  // a(b)
+  '(', PREC_CALL, (node,b) => ( idx++, b=expr(0,CPAREN), idx++,
+    Array.isArray(b) && b[0]===',' ? (b[0]=node, b) : b ? [node, b.valueOf()] : [node]
+  ),
+  // (a+b)
+  '(', PREC_GROUP, (node,b) => !node && (++idx, b=expr(0,CPAREN) || err(), ++idx, b),
+  ')',,,
+]; i < ops.length;) operator(ops[i++],ops[i++],ops[i++])
+
+
+export default parse

package/subscript.js

@@ -1,5 +1,5 @@
-import parse from './src/parse.js'
-import evaluate from './src/evaluate.js'
+import parse from './parse.js'
+import evaluate from './evaluate.js'
 
 export { parse, evaluate }
 

package/subscript.min.js

@@ -1 +1 @@
-var e,r,t,l=l=>(r=l,e=t=0,u()),n=()=>{for(;f()<=32;)e++},o=(r,l,n,o=3)=>{if(e&&t&&t[3]===e)return t;for(;o;)if(null!=(n=r[l=s(o--)]))return t=[l,n,l.length,e]},u=(r,t=-1)=>{n();let l,a,i,c=f(),y=0,A=e;if(c!==r){for(;A===e&&y<p.length;)a=p[y++](c);if(A===e)(l=o(h))&&(e+=l[2],a=[l[0],u(r,l[1])]);else{if(n(),c=f(),c===r)return a;for(y=0;y<d.length;)(i=d[y](a,c))!==a?(a=i,y=0,n(),c=f()):y++}for(n();(c=f())&&c!==r&&(l=o(g))&&l[1]>t;){a=[l[0],a];do{e+=l[2],a.push(u(r,l[1]))}while(s(l[2])===l[0]);n()}return a}},a=e=>(e=c((e=>e>=48&&e<=57||e>=65&&e<=90||e>=97&&e<=122||36==e||95==e||e>=192)))&&y.hasOwnProperty(e)?y[e]:e,f=()=>r.charCodeAt(e),s=(t=1)=>r.substr(e,t),i=r=>{throw Error(r+" at "+e)},c=(t=1,l=e)=>{if("number"==typeof t)e+=t;else for(;t(f());)++e>r.length&&i("Unexpected end "+t);return e>l?r.slice(l,e):null},p=l.token=[(r,t)=>40===r?(e++,t=u(41),e++,t):null,(r,t,l)=>{if(r=c(l=e=>e>=48&&e<=57)||"",46===f()&&(e++,r+="."+c(l)),r)return(69===(t=f())||101===t)&&(e++,r+="e",(43===(t=f())||45===t)&&(r+=s(),e++),r+=c(l)),parseFloat(r)},(r,t)=>34===r?(t=s(),e++,t+c((e=>e!==r))+(e++,t)):null,a],y=l.literal={true:!0,false:!1,null:null},d=l.postfix=[(r,t,l)=>(46===t?(e++,n(),r=[".",r,'"'+a()+'"']):91===t?(e++,r=[".",r,u(93)],e++):40===t&&(e++,l=u(41),r=Array.isArray(l)&&","===l[0]?(l[0]=r,l):null==l?[r]:[r,l],e++),r),(t,l)=>43!==l&&45!==l||r.charCodeAt(e+1)!==l?t:[c(2),t]],h=l.unary={"-":17,"!":17,"+":17,"++":17,"--":17},g=l.binary={",":1,"||":6,"&&":7,"|":8,"^":9,"&":10,"==":11,"!=":11,"<":12,">":12,"<=":12,">=":12,"<<":13,">>":13,">>>":13,"+":14,"-":14,"*":15,"/":15,"%":15},A=l,b=e=>Array.isArray(e)&&("string"==typeof e[0]||b(e[0])),m=(e,r={},t,l)=>b(e)?("string"==typeof(t=e[0])&&(l=v[t]),"function"!=typeof(t=l||m(t,r))?t:t.call(...e.map((e=>m(e,r))))):e&&"string"==typeof e?'"'===e[0]?e.slice(1,-1):"@"===e[0]?e.slice(1):e in r?r[e]:e:e,v=m.operator={"!":e=>!e,"++":e=>++e,"--":e=>--e,".":(...e)=>e.reduce(((e,r)=>e&&e[r])),"%":(...e)=>e.reduce(((e,r)=>e%r)),"/":(...e)=>e.reduce(((e,r)=>e/r)),"*":(...e)=>e.reduce(((e,r)=>e*r)),"+":(...e)=>e.reduce(((e,r)=>e+r)),"-":(...e)=>e.length<2?-e:e.reduce(((e,r)=>e-r)),">>>":(e,r)=>e>>>r,">>":(e,r)=>e>>r,"<<":(e,r)=>e<<r,">=":(e,r)=>e>=r,">":(e,r)=>e>r,"<=":(e,r)=>e<=r,"<":(e,r)=>e<r,"!=":(e,r)=>e!=r,"==":(e,r)=>e==r,"&":(e,r)=>e&r,"^":(e,r)=>e^r,"|":(e,r)=>e|r,"&&":(...e)=>e.every(Boolean),"||":(...e)=>e.some(Boolean),",":(...e)=>e.reduce(((e,r)=>r))},w=m,x=e=>(e="string"==typeof e?A(e):e,r=>w(e,r));export{x as default,w as evaluate,A as parse};
+var e,r,a=15,t=(a,t)=>(r=a,e=0,t=f(),e<r.length?o():t.valueOf()),o=(a="Bad syntax")=>{throw Error(a+" `"+r[e]+"` at "+e)},n=(a=1,t=e)=>{if("number"==typeof a)e+=a;else for(;a(s());)e++;return r.slice(t,e)},s=(a=0)=>r.charCodeAt(e+a),l=(a=1)=>r.substr(e,a),f=(e=0,r,a,n,s=0,l,f)=>{for(;(a=t.space())&&(f=i[a]?.(n,e)||!n&&p(a));)n=f;return r&&a!==r&&o("Unclosed paren"),n},u=(t.space=r=>{for(;(r=s())<=32;)e++;return r},t.token=[e=>(e=n((e=>e>47&&e<58||46==e)))&&((69==s()||101==s())&&(e+=n(2)+n((e=>e>=48&&e<=57))),isNaN(e=new Number(e))?o("Bad number"):e),(e,r)=>34==e&&n()+n((r=>r-e))+n(),e=>n((e=>e>=48&&e<=57||e>=65&&e<=90||e>=97&&e<=122||36==e||95==e||e>=192&&215!=e&&247!=e))]),p=(e,r=0,a)=>{for(;r<u.length;)if(a=u[r++](e))return a},i=[],c=(r,a=0,u=0,p,c=r.charCodeAt(0),y=r.length,g=i[c],h=r.toUpperCase()!==r,v)=>(v=y<2?h?e=>s(1)<=32:e=>1:h?e=>l(y)==r&&s(y)<=32:e=>l(y)==r,p=u?u>0?e=>e&&[n(y),e]:u<0?e=>!e&&[n(y),(f(a-1)||o()).valueOf()]:u:n=>{n=[r,n||o()];do{e+=y,n.push((f(a)||o()).valueOf())}while(t.space()==c&&v());return n},i[c]=(e,r)=>r<a&&v()&&p(e)||g&&g(e,r));for(let r=0,t=[",",1,,"|",6,,"||",4,,"&",8,,"&&",5,,"^",7,,"==",9,,"!=",9,,">",10,,">=",10,,">>",11,,">>>",11,,"<",10,,"<=",10,,"<<",11,,"+",12,,"+",a,-1,"++",a,-1,"++",a,1,"-",12,,"-",a,-1,"--",a,-1,"--",a,1,"!",a,-1,"*",13,,"/",13,,"%",13,,".",18,(e,r)=>e&&[n(),e,"string"==typeof(r=f(18))?'"'+r+'"':r.valueOf()],"[",18,r=>(e++,r=[".",r,f(0,93).valueOf()],e++,r),"]",,,"(",18,(r,a)=>(e++,a=f(0,41),e++,Array.isArray(a)&&","===a[0]?(a[0]=r,a):a?[r,a.valueOf()]:[r]),"(",19,(r,a)=>!r&&(++e,a=f(0,41)||o(),++e,a),")",,,];r<t.length;)c(t[r++],t[r++],t[r++]);var y=t,g=e=>Array.isArray(e)&&("string"==typeof e[0]||g(e[0])),h=(e,r={},a,t)=>g(e)?("string"==typeof(a=e[0])&&(t=d[a]),"function"!=typeof(a=t||h(a,r))?a:a.call(...e.map((e=>h(e,r))))):e&&"string"==typeof e?'"'===e[0]?e.slice(1,-1):"@"===e[0]?e.slice(1):e in r?r[e]:e:e,v=e=>(...r)=>r.reduce(e),d=h.operator={"!":e=>!e,"++":e=>++e,"--":e=>--e,".":v(((e,r)=>e&&e[r])),"%":v(((e,r)=>e%r)),"/":v(((e,r)=>e/r)),"*":v(((e,r)=>e*r)),"+":v(((e,r)=>e+r)),"-":(...e)=>e.length<2?-e:e.reduce(((e,r)=>e-r)),">>>":(e,r)=>e>>>r,">>":(e,r)=>e>>r,"<<":(e,r)=>e<<r,">=":(e,r)=>e>=r,">":(e,r)=>e>r,"<=":(e,r)=>e<=r,"<":(e,r)=>e<r,"!=":(e,r)=>e!=r,"==":(e,r)=>e==r,"&":(e,r)=>e&r,"^":(e,r)=>e^r,"|":(e,r)=>e|r,"&&":(...e)=>e.every(Boolean),"||":(...e)=>e.some(Boolean),",":v(((e,r)=>r))},A=h,O=e=>(e="string"==typeof e?y(e):e,r=>A(e,r));export{O as default,A as evaluate,y as parse};

package/src/parse.js

@@ -1,134 +0,0 @@
-const PERIOD = 46, OPAREN = 40, CPAREN = 41, OBRACK = 91, CBRACK = 93, PLUS = 43, MINUS = 45
-
-export let index, current, lastOp
-
-export const parse = str => (current=str, index=lastOp=0, expr()),
-
-space = () => { while (code() <= 32) index++ },
-
-// consume operator
-operator = (ops, op, prec, l=3) => {
-  // memoize by index - saves 20% to perf
-  if (index && lastOp && lastOp[3] === index) return lastOp
-
-  // ascending lookup is faster for 1-char operators, longer for 2+ char ops, so we use descending
-  while (l) if ((prec=ops[op=char(l--)])!=null) return lastOp = [op, prec, op.length, index] //opinfo
-},
-
-expr = (end, prec=-1) => {
-  space()
-
-  let cc = code(), op, node, i=0, mapped, from=index
-
-  if (cc === end) return
-
-  // parse node by token parsers (direct loop is faster than token.find)
-  while (from===index && i < token.length) node = token[i++](cc)
-
-  // unary prefix
-  if (from===index) (op = operator(unary)) && (index += op[2], node = [op[0], expr(end, op[1])])
-
-  // postfix handlers allow a.b[c](d).e, postfix operators, literals etc.
-  else {
-    if (space(), cc=code(), cc === end) return node
-    for (i=0; i < postfix.length;) if ((mapped=postfix[i](node, cc)) !== node) node=mapped, i=0, space(), cc=code(); else i++
-  }
-  // ALT: seems to be slower
-  // else do {space(), cc=code()} while (postfix.find((parse, mapped) => (mapped = parse(node, cc)) !== node && (node = mapped)))
-
-  space()
-
-  // consume expression for current precedence or higher
-  while ((cc = code()) && cc !== end && (op = operator(binary)) && op[1] > prec) {
-    node = [op[0], node]
-    // consume same-op group, do..while both saves op lookups and space
-    do { index += op[2], node.push(expr(end, op[1])) } while (char(op[2]) === op[0])
-    space()
-  }
-
-  return node;
-},
-
-// tokens
-// 1.2e+3, .5
-float = (number, c, isDigit) => {
-  number = skip(isDigit = c => c >= 48 && c <= 57) || ''
-  if (code() === PERIOD) index++, number += '.' + skip(isDigit)
-  if (number) {
-    if ((c = code()) === 69 || c === 101) { // e, E
-      index++, number += 'e'
-      if ((c=code()) === PLUS || c === MINUS) // +-
-        number += char(), index++
-      number += skip(isDigit)
-    }
-    return parseFloat(number)
-  }
-},
-
-// "a"
-string = (q, qc) => q === 34 ? (qc = char(), index++, qc) + skip(c => c !== q) + (index++, qc) : null,
-
-// (...exp)
-group = (c, node) => c === OPAREN ? (index++, node = expr(CPAREN), index++, node) : null,
-
-// var or literal
-id = name => (name = skip(c =>
-  (c >= 48 && c <= 57) || // 0..9
-  (c >= 65 && c <= 90) || // A...Z
-  (c >= 97 && c <= 122) || // a...z
-  c == 36 || c == 95 || // $, _,
-  c >= 192 // any non-ASCII
-)) && literal.hasOwnProperty(name) ? literal[name] : name,
-
-
-// ------------ util
-code = () => current.charCodeAt(index), // current char code
-char = (n=1) => current.substr(index, n), // skip n chars
-err = (msg) => { throw Error(msg + ' at ' + index) },
-skip = (is=1, from=index) => { // consume N or until condition matches
-  if (typeof is === 'number') index += is
-  else while (is(code())) ++index > current.length && err('Unexpected end ' + is) // 1 + true === 2;
-  return index > from ? current.slice(from, index) : null
-},
-
-
-// ----------- config
-token = parse.token = [ group, float, string, id ],
-
-literal = parse.literal = {true:true, false:false, null:null},
-
-postfix = parse.postfix = [
-  // a.b[c](d), 3 in 1 for performance
-  (node, cc, arg) => {
-    if (cc === PERIOD) index++, space(), node = ['.', node, '"'+id()+'"']
-    else if (cc === OBRACK) index++, node = ['.', node, expr(CBRACK)], index++
-    else if (cc === OPAREN)
-      index++, arg=expr(CPAREN),
-      node = Array.isArray(arg) && arg[0]===',' ? (arg[0]=node, arg) : arg == null ? [node] : [node, arg],
-      index++
-    return node
-  },
-
-  // a++, a--
-  (node, cc) => (cc===0x2b || cc===0x2d) && current.charCodeAt(index+1)===cc ? [skip(2), node] : node,
-],
-
-unary = parse.unary = {
-  '-': 17,
-  '!': 17,
-  '+': 17,
-  '++': 17,
-  '--': 17
-},
-
-binary = parse.binary = {
-  ',': 1,
-  '||': 6, '&&': 7, '|': 8, '^': 9, '&': 10,
-  '==': 11, '!=': 11,
-  '<': 12, '>': 12, '<=': 12, '>=': 12,
-  '<<': 13, '>>': 13, '>>>': 13,
-  '+': 14, '-': 14,
-  '*': 15, '/': 15, '%': 15
-}
-
-export default parse